Conventional Huffman encoding algorithms involve constant sorting and scanning of the data. It is very costly in terms of computation time and hardware real estate.
Moreover, the Huffman encoding algorithm also needs to build and maintain a binary tree as coding. This involves a pointer-chasing process. Pointer chasing is very computation-costly.